Identification of Problematic Drywall: Source Markers and ...Contract Standard Operating Procedure...

DRAFT

IDENTIFICATION OF PROBLEMATIC DRYWALL: SOURCE MARKERS AND DETECTION METHODS

Prepared For:

Lori Saltzman, M.S. Director, Division of Health Sciences

U. S. Consumer Product Safety Commission 4330 East West Highway

Bethesda, MD 20814

Prepared By:

Environmental Health & Engineering, Inc. 117 Fourth Avenue

Needham, MA 02494-2725

May 28, 2010

EH&E Report 16512

This report was prepared for the Commission pursuant to contract CPSC-S-09-0027. It has not been reviewed or approved by, and may not necessarily reflect the views of, the Commission.

DRAFT

TABLE OF CONTENTS

1.0 EXECUTIVE SUMMARY ............................................................................................ 1

1.1 BACKGROUND ..................................................................................................... 1

1.2 OBJECTIVES ........................................................................................................ 2

1.3 METHODS ............................................................................................................ 2

1.4 RESULTS AND CONCLUSIONS .......................................................................... 3

2.0 INTRODUCTION ......................................................................................................... 6

2.1 BACKGROUND ..................................................................................................... 6

2.2 OVERVIEW AND OBJECTIVES ........................................................................... 8

2.3 METHODS SUMMARY ......................................................................................... 8

3.0 SOURCE MARKER ANALYSIS—STRONTIUM ....................................................... 12

3.1 INTRODUCTION ................................................................................................. 12

3.2 METHODS .......................................................................................................... 13

3.3 RESULTS ............................................................................................................ 16

4.0 SOURCE MARKER ANALYSIS—ORTHORHOMBIC SULFUR ............................... 25

4.1 INTRODUCTION ................................................................................................. 25

4.2 METHODS .......................................................................................................... 26

4.3 RESULTS ............................................................................................................ 30

5.0 CHAMBER-BASED CORROSION ............................................................................ 36

5.1 INTRODUCTION ................................................................................................. 36

5.2 METHODS .......................................................................................................... 36

5.3 RESULTS ............................................................................................................ 39

6.0 SOURCE MARKERS AND EFFECT ........................................................................ 41

6.1 SOURCE MARKERS AND GAS EMISSIONS .................................................... 41

6.2 SOURCE MARKERS AND CORROSION .......................................................... 44

7.0 DISCUSSION ............................................................................................................ 49

7.1 OVERVIEW ......................................................................................................... 49

7.2 STRONTIUM CONCENTRATIONS USING PORTABLE XRF ANALYZERS ..... 49

7.3 COMPARISON OF ANALYTICAL METHODS FOR ORTHORHOMBIC SULFUR .............................................................................................................. 50

7.4 INTRA-BOARD VARIABILITY OF STRONTIUM AND ORTHORHOMBIC SULFUR CONCENTRATIONS ........................................................................... 50

7.5 EXPOSURE PATHWAY: SOURCE – EXPOSURE – EFFECT .......................... 51

7.6 SUMMARY .......................................................................................................... 52

8.0 CONCLUSIONS ........................................................................................................ 54

9.0 REFERENCES .......................................................................................................... 56

DRAFT

TABLE OF CONTENTS (Continued)

LIST OF TABLES Table 2.1 List of Drywall Samples and Corresponding Analyses Completed of

Orthorhombic Sulfur and Corrosion Table 3.1 Portable X-ray Fluorescence (XRF) Analyzers Used in this Study Table 3.2 Strontium Intra-board Variability Samples Table 3.3 Correlation Matrix of Strontium Concentrations Measured by Four Different

XRF Analyzers Table 4.1 Overview of Analytic Methods Used to Determine Orthorhombic Sulfur

Concentrations Table 4.2 Description of Samples Used for Intra-board Variability Tests Table 4.3 Comparison of Orthorhombic Sulfur Concentrations (mg/kg) Measured Using

Three Different Methods Table 4.4 Orthorhombic Sulfur Concentrations (mg/kg) at Multiple Locations Per Board Table 5.1 Comparison of Corrosion Rates (A/30d) between Sample and Duplicate

Samples Table 6.1 Regression Model Results Showing Predictors of House Average Hydrogen

Sulfide Concentrations (Natural log-transformed) in Indoor Air Table 6.2 Regression Model Results Showing Predictors of House Average Hydrogen

Sulfide Concentrations (Natural log-transformed) in Indoor Air Table 6.3 Summary Table Table 6.4 Regression Model Results Showing Predictors of Silver Corrosion Rate at the

AHU Air Register (Natural log-transformed) Table 6.5 Regression Model Results Showing Predictors of Copper Corrosion Rate at

the AHU Air Register (Natural log-transformed)

DRAFT


LIST OF FIGURES Figure 3.1 Mean Strontium Concentrations Obtained from Repeat Measurements by

XRF Figure 3.2 Correlation of Strontium Concentration Measured by ICP and Handheld XRF,

Strontium Concentration Ranged Between 0 – 7000 mg/kg Figure 3.3 Correlation of Strontium Concentration Measured by ICP and Handheld XRF,

Excluding the 3 Samples with Strontium Concentration Greater than 5,000 mg/kg

Figure 3.4 Comparison of Strontium Measurements Made on Intact and Crushed Drywall Samples

Figure 3.5 Distribution of Strontium Measurements Made at Twelve Unique Locations on Each Drywall Board

Figure 3.6 Comparison of Strontium Concentrations (mg/kg) and Carbonate (Absorbance) Measured in Drywall Samples from Catalog Set at Time of EH&E’s 51-home Study

Figure 3.7 Comparison of Strontium Concentrations (mg/kg) and Carbonate (Absorbance) Measured in Drywall Samples from Full Catalog Set Used in this Study

Figure 4.1 Comparison of S8 Concentrations Measured Using GC/MS and GC/ECD by Columbia Analytical (n=74)

Figure 4.2 Comparison of S8 Concentrations (mg/kg) in Duplicate Samples as Measured by GC/ECD

Figure 4.3 Comparison of S8 Concentrations Measured in Drywall Samples from Complaint and Non-complaint Homes in the 51-home Study

Figure 5.1 Schematic of the Chamber Testing Equipment Figure 6.1 Scatterplot Showing the Relationship Between a) S8 Concentration (mg/kg)

and Hydrogen Sulfide Flux and b) Strontium (mg/kg) and Hydrogen Sulfide Flux

Figure 6.2 Scatterplot Showing the Relationship between S8 Concentration (mg/kg) and a) Sulfur Dioxide Emission Rate and b) Carbon Disulfide Emission Rate

Figure 6.3 Correlation of S8 and Strontium in a) Catalog Samples and b) Homes in the 51-home Study

Figure 6.4 Comparison of S8 Concentrations (mg/kg) and Chamber-based Corrosion Rates (A/30d)

Figure 6.5 Comparison of Strontium Concentrations (mg/kg) and Chamber-based Corrosion Rates (A/30d)

Figure 6.6 Comparison of House-average S8 Concentrations (mg/kg) and Corrosion Rates at the AHU Air Register from the 51-home Study

Figure 6.7 Scatterplot Comparing Strontium Concentrations (mg/kg) and Silver and Copper Corrosion Rates (A/30 days) at the AHU Air Register

DRAFT


LIST OF ABBREVIATIONS AND ACRONYMS

A/30d angstroms per 30 days Ag2S silver sulfide AgCl silver chloride AHU air handling unit CPSC U. S. Consumer Product Safety Commission CS2 carbon disulfide CuO copper oxide Cu2S copper sulfide DFTPP decafluorotriphenylphosphine DRIFTS diffuse reflectance infrared Fourier transform spectroscopy EH&E Environmental Health & Engineering, Inc. EPA U.S. Environmental Protection Agency ERT Environmental Response Team FTIR Fourier transform infrared spectroscopy GC/MS gas chromatography/mass spectrometry GC/ECD gas chromatography electron capture detector H2O2 hydrogen peroxide H2S hydrogen sulfide HCI hydrochloric acid HNO3 nitric acid ICP inductively coupled plasma ICP-AES inductively coupled plasma-atomic emission spectrometry ISA Instrumentation, Systems, and Automation Society LBNL Lawrence Berkeley National Laboratory LOD limit of detection LPM liters per minute mg/kg milligram per kilogram mL milliliter PTFE polytetrafluoroethylene REAC Response Engineering and Analytical Contract RH relative humidity S8 orthorhombic sulfur SO2 sulfur dioxide SOP standard operating procedure SRM standard reference materials XRF x-ray fluorescence °F degrees Fahrenheit L microliter m micrometer

DRAFT

DRAFT Identification of Problematic Drywall Source Marker Characterization May 28, 2010 Environmental Health & Engineering, Inc.16512 Page 1 of 56

1.0 EXECUTIVE SUMMARY

1.1 BACKGROUND

The U.S. Consumer Product Safety Commission (CPSC) has been investigating

homeowner reports of corrosion and adverse health effects in homes constructed with

problematic wallboard that has also been termed ‘Chinese drywall’. To date, CPSC has

received approximately 3,300 homeowner reports from 37 states. The vast majority of

complaints are from people with homes around the Gulf Coast and coastal Virginia.

CPSC initiated a multi-track investigation to examine health and safety concerns

potentially associated with this building product. As part of this investigation,

Environmental Health & Engineering, Inc. (EH&E) conducted a 51-home study in the

summer of 2009 to characterize the relationship between source markers, gas

concentrations, building dynamics, and corrosion. A complete report outlining the

methods, results and findings of that study are available on-line at the CPSC website.

The results of the 51-home study demonstrated that carbonate and strontium measured

by the combination of Fourier transform infrared spectroscopy (FTIR) and x-ray

fluorescence (XRF), respectively, were reliable indicators of Chinese drywall in these

study homes. These study homes qualified for inclusion in the study by meeting specific

pre-established criteria that included homeowner reports of corrosion, odor, possible

health issues, and whether or not homes were constructed or used drywall for

renovations during the period of interest (2006 – 2007). The utility of the

carbonate/strontium marker determined using FTIR/XRF when applied to homes without

these pre-selection criteria is unknown. Additionally, during the time that the in-home

study was conducted, additional source characterization work by government agencies

and private laboratories identified orthorhombic sulfur (S8) as a potential marker of not

only Chinese drywall, but problematic drywall, defined hereafter as drywall associated

with elevated rates of copper and silver corrosion, emissions of certain reduced sulfur

gases, and possessing a distinctive malodor.

Homeowners, government agencies and other parties have a need to reliably determine

which homes have problematic drywall. While several markers have been proposed, the

DRAFT


robustness of the individual markers used to identify problematic drywall has not been

evaluated extensively. EH&E was contracted by CPSC to conduct a detailed

characterization of source markers of problematic drywall.

1.2 OBJECTIVES

The objective of the source characterization study was to evaluate proposed markers of

problematic drywall, defined as drywall associated with elevated rates of copper and

silver corrosion. In this study, we investigated two proposed markers—strontium and S8.

The robustness of each source marker was evaluated by first comparing different

instruments/methods for quantifying each, examining parameters such as within-board

variability and determining method precision. The source markers were then compared

to both chamber-based and field-based measurements of gases and corrosion.

1.3 METHODS

Drywall samples of varying sizes from thirty-five unique boards were supplied to EH&E

by CPSC for analysis. Drywall samples provided to EH&E by CPSC were collected by

CPSC staff from manufacturers, drywall suppliers, and storage warehouses. North

American drywall samples were manufactured in 2009 while Chinese drywall samples

were manufactured from 2005 through 2009. These drywall boards, identified in this

report as ‘catalog samples’ and labeled by EH&E as CPSC1-CPSC35, are of known

origin and represent a diverse cross-section of domestic and foreign drywall.

Many of the drywall samples sent to EH&E were sub-samples obtained from larger

sheets (source-samples) of drywall board that have been retained by the CPSC. Several

government organizations also received sub-samples from the same larger source-

sample drywall boards. This allowed for comparison of chemical measurement data in

some instances.

In addition to the catalog samples, EH&E also had access to archived samples of

drywall obtained from each room of all homes examined in the 51-home study. During

that field study, a coring tool 1 centimeter in diameter was used in areas behind

electrical outlets/faceplates to obtain a sample of drywall up to approximately 0.5 grams

DRAFT


in mass. Additionally, one larger sample (1’x1’) was also obtained from each home. A

selection of these samples was analyzed for source markers to enable a comparison

with in-home gas and corrosion measurements that were previously obtained.

Drywall samples were analyzed for strontium and S8 source markers using several

different techniques. Strontium concentrations were determined using multiple portable

XRF analyzers as well as inductively-coupled plasma atomic emission spectrometry. S8

concentrations were determined in all samples by gas chromatography electron capture

detector (GC/ECD). A comparison of analytic methods for S8 was also undertaken. In a

subset of samples, S8 was determined using three different methods: GC/ECD, gas

chromatography mass spectrometry (GC/MS) using a toluene-based extraction, and

U.S. Environmental Protection Agency’s (EPA’s) Response Engineering and Analytical

Contract Standard Operating Procedure (REAC SOP1805) (EPA 2009a, EPA 2009b).

Drywall samples were also tested for their potential to cause corrosion. Drywall samples

from the catalog samples provided by CPSC were placed in a sealed chamber for eight

days at a constant temperature and humidity (90 degrees Fahrenheit [°F]; 50% relative

humidity [RH]). A corrosion classification coupon containing pre-cleaned strips of both

silver and copper was added to the chamber with the drywall sample to record rates of

corrosion.

1.4 RESULTS AND CONCLUSIONS

The following observations were made during the course of this investigation:

Strontium is a useful, but non-specific marker of problematic drywall when used in

isolation

Elevated strontium concentrations were observed in all problematic drywall, but

also in some non-problematic drywall. Strontium concentrations were correlated

with orthorhombic sulfur concentrations in problematic drywall. Therefore, in the

51-home study where homes were pre-screened based on specific criteria

contained in a CPSC questionnaire (EH&E, 2010) strontium was found to be

predictive of problematic drywall. Strontium content in drywall measured by XRF

is non-destructive, field portable and nearly instantaneous, and, therefore,

DRAFT


remains a useful marker of problematic drywall when used as part of a multi-level

screening approach.

Analysis of strontium in drywall samples can be reliably performed using XRF

Strontium measurements using XRF were accurate when compared to strontium

as determined by inductively coupled plasma-atomic emission spectrometry

(ICP-AES) (inductively coupled plasma-atomic emission spectroscopy). Models

from three different manufacturers yielded highly similar response factors and

strongly correlated results (1:1 slope). Instrument method calibration specific to

strontium is necessary to ensure accuracy of the measurements.

Orthorhombic sulfur (S8) is a sensitive and specific marker of problematic drywall

Orthorhombic sulfur concentrations in drywall were associated with chamber-

based measurements of hydrogen sulfide and corrosion. Consistent findings

were observed when this relationship was evaluated using archived samples of

drywall and measurements of hydrogen sulfide and corrosion in the 51-home

study.

Orthorhombic sulfur (S8) was not detected in any drywall samples from the non-

complaint homes in the 51-home study

Three drywall samples from each of the 10 non-complaint homes in the 51-home

study were analyzed for orthorhombic sulfur. Orthorhombic sulfur was not

detected in any of these samples. In contrast, orthorhombic sulfur concentrations

in the complaint homes ranged from <5 milligrams per kilogram (mg/kg) to

830 mg/kg (median = 54 mg/kg), and were significantly higher than the levels in

the non-complaint homes.

Orthorhombic sulfur (S8) determined using two toluene-based extraction methods

showed strong agreement

Orthorhombic sulfur concentrations determined using the GC/MS (toluene

extraction) and GC/ECD methods showed excellent agreement. Results using

EPA’s REAC SOP 1805 did not show consistent agreement when compared with

the other two methods, in a limited number of samples. GC/ECD appears to be

DRAFT


an attractive option for future analysis of drywall samples due to its potentially

lower cost for laboratories with this capability.

Orthorhombic sulfur (S8) and strontium both exhibited low intra-board variability

Repeat measurements of orthorhombic sulfur and strontium on different locations

of the same drywall board showed strong consistency.

DRAFT


2.0 INTRODUCTION

2.1 BACKGROUND

The CPSC has been investigating homeowner reports of corrosion and adverse health

effects associated with ‘Chinese drywall’. To date, CPSC has received approximately

3,300 homeowner reports from 37 states. The vast majority of complaints are from

people with homes around the Gulf Coast and coastal Virginia.

CPSC initiated a multi-track investigation to examine health and safety concerns

potentially associated with this building product. As part of this investigation, EH&E

conducted a 51-home study in the summer of 2009 to characterize the relationship

between source markers, gas concentrations, building dynamics, and corrosion. A

complete report outlining the methods, results and findings of that study are available

on-line at the CPSC website. The key findings of the 51-home study are:

Study homes with imported drywall had elevated rates of objectively quantified

corrosion

CPSC complaint homes were found to have elevated rates of corrosion as

measured objectively by metal coupon and visual inspection, compared to non-

complaint homes. This finding remained when house status was determined

using an objective source marker of imported drywall (carbonate and strontium

measured using FTIR and XRF, respectively).

The presence of drywall with the carbonate/strontium objective source marker was

associated with increased levels of hydrogen sulfide in indoor air

Homes with the source marker of imported drywall had significantly greater

hydrogen sulfide concentrations compared to non-complaint homes.

Hydrogen sulfide concentrations in air were associated with higher dew points for

complaint homes

A positive association was observed between elevated dew points and hydrogen

sulfide concentrations for homes with the source marker of imported drywall.

DRAFT


Hydrogen sulfide was present where the dew point reached typical room

temperatures and condensation of water vapor would be expected.

Hydrogen sulfide and formaldehyde concentrations in indoor air were associated with

elevated corrosion rates

Hydrogen sulfide was associated with corrosion rates in these study homes. For

silver, a potential interactive effect was seen with formaldehyde; the effect of

formaldehyde on corrosion rates was dependent upon the presence of hydrogen

sulfide. Formaldehyde may be a marker of some other factor associated with

corrosion (e.g., formic acid).

Objective source markers of imported drywall in CPSC complaint homes can be

quantified using portable FTIR and XRF analyzers

FTIR and XRF analyzers provide additional metrics for characterizing drywall in

homes that can be used in conjunction with objectively measured corrosion rates

and malodor.

Air exchange rates in the study homes were all on the low-end of typical air

exchange rates in homes

Both complaint and non-complaint homes were newly constructed homes with air

exchange rates consistent with the low end of the distribution for North American

housing stock (i.e., tightly constructed homes). These low air exchange rates

may play an important role in the effect of gases and indoor environmental

conditions on corrosion and possible exposures to indoor contaminants.

The results of the 51-home study demonstrated that carbonate and strontium measured

by the combination of FTIR and XRF, respectively, were reliable indicators of Chinese

drywall in these study homes. These study homes qualified for inclusion in the study by

meeting specific pre-established criteria that included homeowner reports of corrosion,

odor, possible health issues and whether or not the home was constructed during the

period of interest (2006 – 2007). The utility of the FTIR/XRF marker when applied to

homes without these pre-selection criteria is unknown. Additionally, during the time that

the in-home study was conducted, additional source characterization work by

government agencies and private laboratories identified S8 as a potential marker of not

DRAFT


only Chinese drywall, but problematic drywall, defined hereafter as drywall associated

with elevated rates of copper and silver corrosion, emissions of certain reduced sulfur

gases, and possessing a distinctive malodor.

2.2 OVERVIEW AND OBJECTIVES

Homeowners, government agencies and other parties have a need to reliably determine

which homes have problematic drywall. While several markers have been proposed, the

robustness of the individual markers used to identify problematic drywall has not been

evaluated extensively. EH&E was contracted by CPSC to conduct a detailed

characterization of source markers of problematic drywall.

This report was prepared by EH&E on behalf of CPSC and describes an investigation of

source markers of problematic drywall. The investigation included:

1. Determination of the precision and accuracy of strontium concentration

measurements in drywall made with field portable instruments.

2. Determination of orthorhombic sulfur content in drywall samples from the CPSC

inventory (‘catalog drywall samples’) and drywall samples archived from the 51-home

study.

3. Characterization of corrosion potential of catalog drywall samples.

4. Identification of source markers of problematic drywall by comparison of source

marker concentrations to both chamber-based and in-home measurements of gases

and corrosion.

2.3 METHODS SUMMARY

Drywall samples of varying sizes from thirty-five unique boards were supplied to EH&E

by CPSC for analysis. These drywall boards, identified in this report as ‘catalog samples’

and labeled by EH&E as CPSC1 – CPSC35, are of known origin and represent a diverse

cross-section of domestic and foreign drywall. Drywall samples were analyzed for source

markers using several different techniques, as well as for the potential to cause

corrosion in chamber tests. A list of the drywall samples and types of analyses

DRAFT


performed on each are outlined in Table 2.1. Detailed descriptions of the analyses

performed can be found in the following sections.

Table 2.1 List of Drywall Samples and Corresponding Analyses Completed for Orthorhombic Sulfur, Strontium and

Corrosion

Catalog ID CPSC ID

S8 Laboratory Method Comparison

S8 Intra-Board

Variability Strontium

Strontium Intra-Board Variability

Chamber Corrosion

Test REAC GC/MS GC/ECD GC/ECD XRF ICP XRF

CPSC1 09-302-1429-02 X X X X X X CPSC2 09-840-9139-05 X X X X CPSC3 09-302-1379-09 X X X X X X X CPSC4 09-840-9858-01 X X X X X X CPSC5 09-810-7932-05 X X X X X CPSC6 09-810-7639-06 X X X CPSC7 09-840-9961-03 X X X CPSC8 09-840-9962-08 X X X X CPSC9 09-810-8213-02 X X X X CPSC10 09-810-7069-06 X X X X X X X CPSC11 09-810-8235-03 X X X CPSC12 09-810-8036-05 X X X CPSC13 09-810-8037-01 X X X CPSC14 09-810-8236-07 X X X X X X CPSC15 09-840-9672-07 X X X X X CPSC16 09-302-2636-03 X X X CPSC17 09-840-9707-05 X X X X X X CPSC18 09-840-9673-08 X X X CPSC19 09-302-1487-02 X X X X X X X CPSC20 09-302-2634-01a X X X X CPSC21 09-302-1492-02 X X X X CPSC22 09-302-1493-02a X X X X X CPSC23 09-302-2631-02b X X X X X X X CPSC24 09-810-7077-02 X X X X X X X CPSC25 09-810-7078-05 X X X X X X CPSC26 09-302-2632-01 X X X X X X CPSC27 09-302-2633-02 X X X X CPSC28 09-302-2635-02 X X X CPSC29 09-840-9667-01 X X X X CPSC30 09-302-2637-02a X X X X CPSC31 09-302-1484-02a X X X X CPSC32 09-840-9175-05 X X X X X X CPSC33 09-840-9174-01 X X X X CPSC34 09-810-7339-10 X X X X X X X CPSC35 09-810-8357-01 X X X X X X Archived samples from the 51-home study

3 samples per home (n=153)

X X

CPSC U.S. Consumer Product Safety Commission S8 orthorhombic sulfur REAC Response Engineering and Analytical Contract GC/MS gas chromatography mass spectrometry GC/ECD gas chromatography/electron capture detection XRF x-ray fluorescence ICP inductively coupled plasma-emission spectroscopy

Many of the drywall samples sent to EH&E were sub-samples obtained from larger

sheets (source-samples) of drywall board that have been retained by the CPSC. Several

DRAFT


government organizations also received sub-samples from the same larger source-

sample drywall boards. This allowed for comparison of chemical measurement data in

some instances.

In addition to the catalog samples, EH&E also had access to archived samples of

drywall obtained from each room of all homes in the 51-home study. During that field

study, a coring tool 1 centimeter in diameter was used in areas behind electrical

outlets/faceplates to obtain a sample of drywall up to approximately 0.5 grams in mass.

Additionally, one larger sample (1’x1’) was also obtained from each home. A selection of

these samples was analyzed for source markers to enable a comparison with in-home

gas and corrosion measurements that were previously obtained.

2.3.1 Statistical Analysis

Statistical analysis included compiling descriptive statistics, scatter plots and box plots.

Box plots depict the mean, median, 10th, 25th, 75th, and 90th percentiles, as well as

individual points beyond the 10th and 90th percentiles. Bivariate statistical relationships

were assessed using Spearman correlation and linear regression. Multiple linear

regression models were used for analysis of variance and multivariate regression. When

normality assumptions were not satisfied for the outcome variable, the variable was

natural log-transformed. Covariate selection for multivariate models was determined

based on modeling results reported in the 51-home study (EH&E 2010) to ensure

comparability between reports. Values below the laboratory reporting limit were

substituted using one-half of the reporting limit in statistical analyses. Statistical

significance for all analyses was defined at the =0.05 level. All statistical analyses of

the study data were performed using SAS statistical software, version 9.1 (Cary, North

Carolina).

2.3.2 Quality Assurance/Quality Control

All sampling was conducted at the EH&E laboratory. XRF analyzers were calibrated for

strontium by the manufacturers prior to use in this study. In addition, internal instrument

background checks were run on each Innov-X instrument in accordance with

DRAFT


manufacturer instructions. Two standard reference materials (SRM) were used for the

strontium testing:

SRM 2709—San Joaquin Soil (Baseline Trace Element Concentration)

This SRM is intended primarily for use in the analysis of soils, sediments, or other

materials of similar matrix.

SRM 2702—Inorganics in Marine Sediment (Baltimore Harbor, Baltimore, Maryland)

This SRM is intended for use in evaluating analytical methods for the determination

of selected elements in marine or fresh water sediment and similar matrices.

Sample logs were used to record the drywall identification number and sample type

during testing. Data files were downloaded daily and saved on EH&E’s central file

server. Estimates of accuracy and precision were an objective of this project and are

discussed in detail in the results sections for strontium and S8.

DRAFT


3.0 SOURCE MARKER ANALYSIS—STRONTIUM

3.1 INTRODUCTION

Evaluating strontium as a reliable marker of problematic drywall included two specific

aims. The first was to determine the precision and accuracy of strontium concentration

measurements made using portable XRF analyzers. The second was to evaluate the

sensitivity and specificity of strontium as a marker of problematic drywall. This section of

the report, Section 3, describes the testing done to address the first aim related to

evaluating strontium measurements. The second aim, which evaluates two source

markers, strontium and sulfur, and their relationships to gases and corrosion is

discussed in Section 6.

Strontium, as measured using portable XRF analyzers, has been proposed as a useful

marker for identifying Chinese Drywall (EPA 2009c, EH&E 2010). In EH&E’s 51-home

field study, homes that contained wallboard with a carbonate and strontium marker

(carbonate measured by FTIR and strontium measured by XRF) were found to have

significantly higher hydrogen sulfide concentrations and corrosion than homes that did

not have this marker present. The decision to use a two-component marker was based

on sampling conducted on a limited number of drywall samples of known origin. That

sampling indicated that using either carbonate or strontium alone could lead to false

positives. The combined carbonate/strontium marker using FTIR/XRF was found to be

sufficient to detect differences in these sets of homes and important for establishing an

empirical relationship between source materials in homes and effects. The extent to

which carbonate and strontium are both needed is more fully explored in this analysis

based on additional drywall samples supplied by CPSC.

The FTIR/XRF method previously developed by CPSC and EH&E to identify suspect

drywall was performed using A2 Technologies brand portable FTIR analyzer and

Innov-X brand portable XRF analyzers (EH&E 2010). CPSC desired to know if portable

XRF instruments from different manufacturers could be used reliably. Additionally, the

overall accuracy and precision of the strontium measurements as determined by XRF

analyzers needed to be more fully characterized before final recommendation of a

sampling protocol can be made.

DRAFT


Therefore, the objectives of this task were:

Objective 1

Compare the agreement of strontium measurements within and among different

brands of portable XRF analyzers.

Objective 2

Evaluate the accuracy of strontium concentrations measured using XRF against total

strontium content in wallboard using inductively-coupled plasma atomic emission

spectrometry for comparison.

Objective 3

Determine the intra-board variability of strontium concentrations and evaluate two

sampling techniques.

Objective 4

Review the two-component carbonate and strontium marker (‘FTIR/XRF marker’) for

identification of problematic wallboard.

3.2 METHODS

3.2.1 X-Ray Fluorescence Analysis

Portable XRF analyzers were used to measure the concentration of strontium in the

catalog samples. XRF analyzers have been used for many years for non-destructive

testing of products and materials, most notably for detecting lead in paint. Briefly, XRF

analyzers use low-energy x-rays to produce high-energy photons to excite electrons in a

sample. Upon excitation, outer shell electrons are replaced by inner orbital electrons.

This generates a fluorescent signature that is unique to each element. A detector on the

instrument analyzes the fluorescence patterns and quantifies the concentration of each

element present in the sample.

Four different XRF analyzers, representing three different brands, were evaluated in this

study (Table 3.1).

DRAFT


Table 3.1 Portable X-ray Fluorescence (XRF) Analyzers Used in this Study

Brand Address Model Analysis Mode Bruker Kennewick, WA TRACER turboSD Mining Elements Innov-X Woburn, MA Alpha Series Soil Thermo Fisher Billerica, MA NitonXL3T500 Mining Thermo Fisher Billerica, MA NitonXL3T900SHE Mining

3.2.2 Sample Testing

3.2.2.1 Method Precision

Intra-instrument measurement precision was determined for three different brands (four

total models) of portable analyzer by conducting repeat sampling at a specific location

on each of the eight boards. For each model/brand, 10 consecutive measurements were

obtained at the specified location on the first board. The process was then repeated for

the remaining seven different drywall boards and two SRMs (10 total drywall boards).

Inter-instrument precision was determined by comparing co-located strontium

measurements made using each XRF analyzer on 35 unique drywall samples. One

location on each drywall sample was identified and marked. To eliminate any

interference and attenuation from surface materials, the paper was removed and each

analyzer/method was used to record measurements on the same location of each

drywall sample. Measurements were taken directly on the core of intact drywall material

(i.e., paper removed) and lasted 30 seconds each.

3.2.2.2 Method Accuracy

To assess accuracy of strontium measurements made using portable XRF analyzers,

drywall samples were obtained from 17 boards for analysis of total strontium

concentration by ICP-AES. Approximately five grams of core material was removed from

each drywall board (i.e., paper removed), crushed and placed in a sampling container.

The sample was then analyzed by handheld XRF analyzer before being shipped to an

analytic laboratory for determination of strontium content by ICP-AES.

The ICP-AES analysis was performed by Columbia Analytical Sciences (Simi,

California). Drywall samples were digested according to EPA Method SW846 3050B,

DRAFT


“Acid Digestion of Sediments, Sludges, and Soils.” One-gram equivalent of the sample

was digested with repeated additions of nitric acid (HNO3) and hydrogen peroxide

(H2O2). Hydrochloric acid (HCl) was added to the initial digestate and the sample was

refluxed prior to dilution to a final volume of 100 milliliters (mL). The digestate was then

analyzed following EPA Method SW846 6010B for strontium using ICP-AES. The

instrument measured characteristic emission spectra by optical spectrometry. The

intensity of emission lines was monitored. Final strontium concentrations were calculated

using information regarding the digestion process and results from the ICP analysis.

3.2.2.3 Intra-Board Variability

Thirteen drywall boards were selected for testing in multiple locations to assess intra-

board variability (Table 3.2). Each drywall board was evenly divided by three columns

and four rows for a total of 12 sampling locations per board. Strontium concentrations

were determined at each of the 12 locations by portable XRF analyzers.

Table 3.2 Strontium Intra-board Variability Samples

Catalog ID Original Sample Size Number of Samples CPSC3 11” x 12” 12 CPSC8 9” x 12” 12 CPSC14 9” x 6” 12 CPSC19 8” x 16” 12 CPSC20 12” x 11” 12 CPSC21 8” x 16” 12 CPSC23 7” x 14” 12 CPSC24 11” x 15” 12 CPSC25 10” x 16” 12 CPSC29 1’ x 1’ 12 CPSC30 1’ x 1’ 12 CPSC32 1’ x 1’ 12 CPSC34 1’ x 1’ 12

3.2.2.4 Fourier Transform Infrared (FTIR) Analysis

All samples tested for strontium by XRF were also tested for carbonate using FTIR. The

analytic method has been previously described (EH&E 2010). Briefly, FTIR

measurements were obtained using the A2 Technologies Exoscan instrument, a full

scanning Fourier transform mid-infrared spectrometer. The diffuse reflectance infrared

DRAFT


Fourier transform spectroscopy (DRIFTS) technique was used in this study to obtain

measurements of relative carbonate content in each sample.

3.3 RESULTS

3.3.1 Objective 1—Precision of Strontium Measurements Across XRF Instruments

3.3.1.1 Intra-Instrument Precision

Results from the repeat strontium measurement testing of one location on eight drywall

boards and two SRMs by XRF (Innov-X brand) are presented in Figure 3.1 and show

high precision. Data points represent the average of 10 measurements and the error

bars represent two standard deviations. Similar results were observed for the

Thermo/Niton and Bruker brand XRF analyzers (not shown). The coefficient of variation

among the 10 repeated measurements across all 40 combinations of instruments (4)

and boards or SRMs (10) ranged from 0.4 – 14% (median=2%). These results

demonstrate strong within-instrument precision for strontium measurements made by

XRF analyzers calibrated for strontium.

DRAFT


Figure 3.1 Mean Strontium Concentrations Obtained from Repeat Measurements by XRF (Innov-X brand analyzer) (Error bars represent 2*standard deviation)

3.3.1.2 Inter-Instrument Precision

The strontium measurements on the core of intact drywall samples obtained using the

four XRF analyzer models/brands were all highly correlated (0.99, p<0.0001) (Table 3.3).

This strong agreement indicates that all of the XRF analyzer models tested perform

similarly with respect to identifying and quantifying the relative amounts of strontium in

drywall. The overall accuracy of each analyzer is discussed in the following section.

NIS

T27

02

CP

SC

6

NIS

T27

09

CP

SC

4

CP

SC

7

CP

SC

15

CP

SC

34

CP

SC

10

CP

SC

23

CP

SC

32

Str

on

tiu

m C

on

cen

trat

ion

(m

g/k

g)

0

1000

2000

3000

4000

5000

6000

CV Range: 1.4 - 5.6%

DRAFT


Table 3.3 Correlation Matrix of Strontium Concentrations Measured by Four Different XRF

Analyzers

Thermo

(Niton XL3T500)

Thermo (Niton

XL3T900SHE) Innov-X

(Alpha Series)

Bruker (TRACER turboSD)

Thermo (NitonXL3T500)

1

Thermo (NitonXL3T900SHE)

0.998 <.0001

1

Innov-X (Alpha Series)

0.997 <.0001

0.998 <.0001

1

Bruker (TRACER turboSD)

0.987 <.0001

0.988 <.0001

0.990 <.0001

1

3.3.2 Objective 2—Accuracy of Strontium Measurements Across XRF Instruments

The accuracy of strontium measurements obtained by XRF was assessed by

comparison to corresponding strontium results measured by ICP-AES. Scatter plots

comparing the measurements made by XRF and ICP-AES are depicted in Figure 3.2.

The XRF and ICP-AES measurement results show excellent agreement for strontium

(slope = 0.85 – 0.95, p<0.01).

DRAFT


Figure 3.2 Correlation of Strontium Concentration Measured by ICP and Handheld XRF, Strontium Concentration Ranged Between 0 – 7000 mg/kg (The solid line represents the 1:1 slope.)

The data depicted in Figure 3.2 show that the slopes are influenced by measurements of

three boards in the 6,000 mg/kg range. It is important to consider the meaningful range

of analysis when analyzing comparative data. For strontium, levels in the range of

1,200 mg/kg have been proposed as the cut point for a marker of problematic drywall.

Therefore, the accuracy of measurements that vastly exceed the range of interest (e.g.,

5,000 mg/kg) may be less important because these values are clearly above reported

thresholds of interest. We re-analyzed the data restricting the data to a range more

relevant to these thresholds. When the data are re-examined restricting the data to

Strontium Concentration by ICP-AES (mg/kg)

0 1000 2000 3000 4000 5000 6000 7000

Str

ontiu

m C

once

ntra

tion

by

XR

F (

mg/

kg)

0

1000

2000

3000

4000

5000

6000

7000


0 1000 2000 3000 4000 5000 6000 7000

Str

ontiu

m C

once

ntra

tion

by X

RF

(m

g/kg

)

0

1000

2000

3000

4000

5000

6000

7000

N = 35Slope = 0.85R-square = 0.96p-value < 0.01



0 1000 2000 3000 4000 5000 6000 7000

Str

ontiu

m C

once

ntra

tion

by X

RF

(m

g/kg

)

0

1000

2000

3000

4000

5000

6000

7000


0 1000 2000 3000 4000 5000 6000 7000

Str

ontiu

m C

once

ntra

tion

by X

RF

(m

g/kg

)

0

1000

2000

3000

4000

5000

6000

7000






Bruker(TRACER turboSD)

DRAFT


values <5000 mg/kg, the accuracy of measurements made by XRF is even greater

(slope = 0.98 – 1.04, p<0.01) (Figure 3.3).

Figure 3.3 Correlation of Strontium Concentration Measured by ICP and Handheld XRF,

Excluding the 3 Samples with Strontium Concentration Greater than 5,000 mg/kg (The solid line represents the 1:1 slope.)

3.3.2.1 Sampling Method—Crushed v. Intact

Analysis of drywall in homes using portable XRF analysis can be done three different

ways. The first is to obtain measurements directly on the drywall in situ through

paint/plaster, the second is to obtain measurements in situ but with the paper and

paint/plaster removed, and the third is to first remove a sample of drywall with a coring

tool or saw, and then analyze the sample. For the first method, we previously reported


0 1000 2000 3000 4000

Str

ontiu

m C

once

ntra

tion

by

XR

F (

mg/

kg)

0

1000

2000

3000

4000


0 1000 2000 3000 4000

Str

ontiu

m C

once

ntra

tion

by

XR

F (

mg/

kg)

0

1000

2000

3000

4000




0 1000 2000 3000 4000

Str

ontiu

m C

once

ntra

tion

by

XR

F (

mg/

kg)

0

1000

2000

3000

4000


0 1000 2000 3000 4000

Str

ontiu

m C

once

ntra

tion

by

XR

F (

mg/

kg)

0

1000

2000

3000

4000

N = 35Slope =1.00R-square = 0.99p-value < 0.01





Bruker(TRACER turboSD)

DRAFT


on the impact of surface coatings on obtaining accurate measurements of the drywall

core using XRF (EH&E 2010). The second and third methods involve measuring the

drywall core without interference from paint/plaster, but the latter method may provide a

sample that is crushed and homogenized as opposed to intact, especially when using a

coring tool. Therefore, we assessed agreement between measurements made directly

on an intact core and then measurements made on the same sample after it was

crushed. The results of this analysis are presented in Figure 3.4. There was a strong,

positive, 1:1 relationship between strontium concentrations measured on an intact core

and measurements made on a 2.5 gram, crushed sample (Spearman r = 0.99,

p<0.0001).

Figure 3.4 Comparison of Strontium Measurements Made on Intact and Crushed Drywall Samples

Strontium Concentration (mg/kg) - Intact Core

0 1000 2000 3000 4000 5000 6000

Str

ontiu

m C

once

ntra

tion

(mg/

kg)

- C

rush

ed S

ampl

e (2

-5 g

ram

s)

0

1000

2000

3000

4000

5000

6000


DRAFT


3.3.3 Objective 3—Assessment of Strontium Intra-Board Variability and Sampling Methods

Results from strontium measurements made using an XRF analyzer (Innov-X) on twelve

locations on each of the ten different drywall boards are presented in Figure 3.5. Similar

results were observed for the Bruker and two Thermo/Niton brand XRF analyzers. Each

box represents variability of the 12 measurements made from each of the 13 respective

drywall boards. The results demonstrate intra-board variability is small for boards with

strontium measurements nominally less than 3,000 mg/kg (coefficient of variation =

2 – 9%). The intra-board measurements demonstrated greater variation at

concentrations greater than 3,000 mg/kg, although the overall precision in this range

may not be as relevant.

Figure 3.5 Distribution of Strontium Measurements Made at Twelve Unique Locations on Each Drywall Board

CP

SC

30

CP

SC

21

CP

SC

14

CP

SC

24

CP

SC

8

CP

SC

20

CP

SC

19

CP

SC

3

CP

SC

25

CP

SC

34

CP

SC

29

CP

SC

23

CP

SC

32

Str

ontiu

m C

once

ntra

tion

(mg/

kg)

Mea

sure

d at

12

Loca

tions

0

1000

2000

3000

4000

5000

6000

DRAFT


3.3.4 Objective 4—Review of the Two-Component Carbonate/Strontium Marker Using FTIR/XRF

During EH&E’s 51-home study, samples of known origin (‘catalog samples’) were

measured to determine if there were unique markers of Chinese drywall. Based on

testing of the catalog samples, it was observed that neither strontium nor carbonate

alone were both sensitive and specific as a marker of “Chinese” drywall (i.e., testing by

either method alone could produce some false positives and false negatives) (Figure

3.6). However, when strontium and carbonate were used in conjunction for analysis of

samples, they were found to be a reliable marker of problematic drywall for that study.

Figure 3.6 Comparison of Strontium Concentrations (mg/kg) and Carbonate (Absorbance)

Measured in Drywall Samples from Catalog Set at Time of EH&E’s 51-home Study

As a result, the 51-home study characterized homes based on whether or not drywall in

the homes had the carbonate/strontium marker (‘FTIR/XRF marker’). This marker was

found to be a significant predictor of hydrogen sulfide and elevated rates of corrosion in

these study homes—homes which were qualified for the study based on their year of

construction and self-reported occupant complaints.

Strontium (mg/kg) - XRF

0 1000 2000 3000 4000 5000 6000

Ca

lciu

m C

arb

onat

e (a

bsor

banc

e) -

FT

IR

0

5

10

15

20

North AmericaChina

DRAFT


Carbonate and strontium results obtained on the full set of drywall samples of known

origin (‘catalog samples’, identified in Table 3.2) are presented in Figure 3.7. When

analyzing a more complete dataset, carbonate and strontium were found to be

significantly and positively correlated, on average (Spearman r=0.72, p<0.01). However,

the potential for false positives remained if only one marker was used.

Figure 3.7 Comparison of Strontium Concentrations (mg/kg) and Carbonate (Absorbance) Measured in Drywall Samples from Full Catalog Set Used in this Study

The second aim of this task was to compare strontium concentrations to gas emissions

and corrosion. This assessment of strontium concentrations as a marker of gases and

corrosion is presented in Section 6.2, Source Markers and Corrosion.

Strontium (mg/kg) - XRF

0 1000 2000 3000 4000 5000 6000

Ca

lciu

m C

arb

ona

te (

ab

sorb

anc

e)

- F

TIR

0

5

10

15

20

North AmericaNorth America - Used Imported MaterialsChina

DRAFT


4.0 SOURCE MARKER ANALYSIS—ORTHORHOMBIC SULFUR

4.1 INTRODUCTION

Similar to the investigation of strontium as a source marker, there were two specific aims

in evaluating orthorhombic sulfur as a marker of problematic drywall. The first aim was to

evaluate the precision and accuracy of orthorhombic sulfur measurements. The second

aim was to evaluate the sensitivity and specificity of orthorhombic sulfur as a marker of

problematic drywall. This section focuses on the first aim—testing done to compare

laboratory methods and evaluate the precision of orthorhombic sulfur measurements.

The analysis of orthorhombic sulfur as a sensitive and specific marker of problematic

drywall through comparison to gas emissions and corrosion rates is described in

Section 6.

Elemental sulfur exists as three allotropes with the most stable and common being S8.

S8 has been proposed as a marker of suspect drywall by the EPA and other

governmental and non-governmental groups.

EH&E conducted an evaluation of S8 that included the following objectives:

Objective 1

Compare three analytical methods (EPA Environmental Response Team

[ERT]/REAC SOP 1805, GC/MS [toluene-extraction] and GC/ECD) for analyzing

orthorhombic sulfur in a subset of drywall samples from the CPSC ‘catalog’ drywall

samples in this study

Objective 2

Assess intra-board variability to determine representativeness of using a single

‘core’ sample to characterize larger pieces of drywall

Objective 3

Determine orthorhombic sulfur concentrations in catalog drywall samples and

archived drywall samples from the 51-home study

DRAFT


4.2 METHODS

4.2.1 Study Design

4.2.1.1 Laboratory Method Comparison—Catalog Samples

To evaluate agreement among the three analytic methods, a subset (n=13) of the

catalog samples were analyzed using each of the three methods discussed previously

(Table 4.1). A 15g sample was removed from each board, homogenized, and divided

into three equivalent subsamples to ensure sample uniformity across methods.

Table 4.1 Overview of Analytic Methods Used to Determine Orthorhombic Sulfur

Concentrations

Analytic Method Description Number of Primary

Samples EPA ERT/REAC SOP 1805 Soxhlet extraction

(dichloromethane:acetone) GC/MS

13

GC/MS (toluene extraction) Solvent extraction (toluene) GC/MS

13

GC/ECD Solvent extraction (toluene) GC/ECD

35

EPA U.S. Environmental Protection Agency ERT/REAC Environmental Response Team Response Engineering and Analytical Contract SOP standard operating procedure GC/MS gas chromatography mass spectrometry GC/ECD gas chromatography electron capture detector

Thirteen drywall samples were selected from the catalog samples for analysis by EPA’s

ERT/REAC SOP 1805 (‘REAC SOP 1805’) and a similar laboratory method that also

uses GC/MS but with a toluene based extraction (‘GC/MS (toluene extraction)’). While

both methods rely on quantification of S8 using GC/MS, the methods differ in the solvent

extraction method and chromatographic column used to separate constituents of the

sample. A complete description of the laboratory methods can be found in the Section

4.2.2.

In addition to the thirteen samples analyzed by both EPA REAC SOP 1805 and the

GC/MS (toluene extraction) method, all of the catalog samples (n = 35) were analyzed

for S8 using a third method—GC/ECD. This method has been proposed as equivalent to

DRAFT


the two methods that use GC/MS for determining S8 concentrations (Singhvi et al. 2009).

Duplicate samples were prepared from six of the catalog boards and analyzed by

GC/ECD to evaluate method precision.

4.2.1.2 Intra-Board Variability

Intra-board variability of S8 was assessed by analyzing multiple samples (n=3 to12) from

eight separate catalog samples (Table 4.2). The boards were marked in evenly

distributed sections and samples were taken from the mid-point of each location. Before

sending the samples for S8 analysis, the samples were measured for strontium

concentration by XRF.

Table 4.2 Description of Samples Used for Intra-board Variability Tests

Catalog ID Original Board Size Number of Samples CPSC 3 1’ x 1’ 12 CPSC 10 9” x 6” 3 CPSC 14 9” x 6” 12 CPSC 15 1’ x 1’ 3 CPSC 25 9.5” x 16” 3 CPSC 32 1’ x 1’ 12 CPSC 34 1’ x 1’ 5 CPSC 35 1’ x 1’ 5 CPSC U.S. Consumer Product Safety Commission

4.2.1.3 Archived In-Home Samples

In addition to the catalog samples, archived samples from EH&E’s 51-home study were

analyzed for S8 concentrations by GC/ECD. During the 51-home study, samples were

cored from the wallboard around electrical outlets in most rooms. These samples were

typically <0.5 grams per core sample. Additionally, a larger piece (~1’x1’) of drywall was

obtained from each home and archived. The cored samples, not originally collected for

this purpose, did not provide sufficient mass to provide detection limits in the

5 – 10 mg/kg range and therefore be comparable to the sensitivity anticipated for

categorizing problematic drywall. Therefore, cored samples from the same room were

composited to provide sufficient mass for analysis and provide a limit of detection (LOD)

in the 5 mg/kg range. For these core composite samples, an equal mass from each of

DRAFT


the individual core samples was combined to evenly represent the different wall boards

sampled from a room. For the large wallboard piece, a 2 – 5 gram sample was removed

and sent for analysis. The result was that each home had up to three total samples

analyzed for S8 – two composite samples (one from each of two rooms), and one non-

composite sample (n=153).

4.2.2 Laboratory Methods

Orthorhombic sulfur analysis was conducted by Columbia Analytical Services (Simi,

California). The laboratory was provided an EH&E sample ID but were blinded to the

origin of all samples. Three methods were used: EPA REAC SOP 1805, GC/MS (toluene

extraction) and GC/ECD. Drywall samples were crushed by EH&E, homogenized, and

divided into three equal subsamples to ensure uniformity. A negative control sample

(known unaffected drywall) was prepared by Columbia Analytical Services and analyzed

along with the various subsamples as an added check for quality control. A summary of

each method was provided by Columbia Analytical Services and each is presented in

the following sections.

4.2.2.1 EPA REAC SOP 1805

Subsamples were extracted using EPA Method SW846 3541, “Automated Soxhlet

Extraction” (Soxtherm). Samples were aliquoted, mixed with sodium sulfate to remove

moisture, and surrogates (2-fluorophenol; phenol-d6; nitrobenzene-d5; 2-fluorobiphenyl;

2,4,6-tribromophenol; and terphenyl-d14) were added to evaluate extraction efficiency.

Samples were then extracted per the method using a 4:1 mixture of

dichloromethane:acetone. The samples were concentrated to a final volume of 1 mL and

solvent exchanged into dichloromethane on an N-EVAP evaporator unit under nitrogen.

All samples contained sediment and therefore were filtered using a 0.45 micrometer

(m) polytetrafluoroethylene (PTFE) syringe tip filter. The extracted samples were

colored, which ranged from light yellow to dark brown. The darkest extract was analyzed

at a 10 fold dilution.

A four point calibration was performed for the surrogate compounds. An EPA Method

SW846 8270 internal standard mixture was added to an aliquot of the extracts. Sample

DRAFT


extracts were analyzed on a GC/MS utilizing a DB-5 column (30m x 0.25mm x 0.25m

film thickness). Helium was used as the carrier gas in the analytical system. S8

allotropes were identified based on the spectral match, comparing the mass spectra of

the sample peak with mass spectra in a comprehensive mass spectral library. The

concentration of S8 was estimated using a response factor of one and the response of

the associated internal standard (phenanthrene-d10). The results are reported as

S8 in mg/kg.

4.2.2.2 GC/MS (Toluene Extraction)

One gram (1g) aliquots of each subsample was solvent extracted with agitation for two

minutes using toluene (5 mL). A 1.0 microliter (L) aliquot of the sample extract is

injected into the gas chromatograph by splitless injection where a fused silica capillary

column separates S8 from other species in the sample and a mass selective detector

operated in the SCAN mode detected the S8. Helium was used as the carrier gas in the

analytical system. The retention time and select characteristic ions of S8 were used for

identification. Quantitative analysis was performed by using an internal standard

calibration procedure, which involves the comparison of instrument responses from the

target compounds in the sample to the response of the internal standard that is added to

the sample prior to analysis. The ratio of the peak area of the target compound in the

sample to the peak area of the internal standard in the sample was compared to a

similar ratio derived for each calibration standard.

Additional instrument quality control checks included daily tuning of the mass

spectrometer using decafluorotriphenylphosphine (DFTPP), a five point calibration for

S8, initial calibration verification standard analysis, and evaluation of extraction surrogate

recovery.

The concentration of S8 in the sample was reported in mg/kg, and was calculated using

the analytical result, the sample weight and the final extract volume.

4.2.2.3 GC/ECD

One gram (1g) aliquots of the subsamples were solvent extracted with agitation for two

minutes using toluene (5 mL). A 1.0 L aliquot of the sample extract was injected into

DRAFT


the gas chromatograph by splitless injection where a fused silica capillary column

separates S8 from other species and an ECD detects the S8. Helium was used as the

carrier gas in the analytical system. The identification of S8 was performed by comparing

the retention time of S8 with the respective retention time of an authentic standard.

Quantitative analysis was performed by using an internal standard calibration procedure,

which involves the comparison of instrument responses from the target compounds in

the sample to the response of the internal standard that is added to the sample prior to

analysis. The ratio of the peak area of the target compound in the sample to the peak

area of the internal standard in the sample was compared to a similar ratio derived for

each calibration standard.

Additional instrument quality control checks included a five point calibration for S8, initial

calibration verification standard analysis, and evaluation of extraction surrogate

recovery. The concentration of S8 in the sample was reported in mg/kg, and was

calculated using the analytical result, the sample weight and the final extract volume.

4.3 RESULTS

4.3.1 Objective 1—Analytic Method Comparison

Results of the S8 measurements of the catalog samples by three different methods are

presented in Table 4.3.

DRAFT


Table 4.3 Comparison of Orthorhombic Sulfur Concentrations (mg/kg) Measured Using

Three Different Methods

Catalog ID REAC SOP 1805 GC/MS

(toluene extraction) GC/ECD CPSC1 ND < 5 < 5 CPSC3 2.3 79 91 CPSC4 ND < 5 < 5 CPSC5 ND < 5 < 5 CPSC10 4.4 6 7.7 CPSC17 ND < 5 < 5 CPSC19 ND < 5 < 5 CPSC22 ND < 5 < 5 CPSC23 ND < 5 < 5 CPSC24 ND < 5 < 5 CPSC26 ND < 5 < 5 CPSC34 650 610 870 CPSC35 64 1,000 1,200 mg/kg milligrams per kilogram REAC REAC SOP 1805 GC/MS gas chromatography mass spectrometry GC/ECD gas chromatography electron capture detector Laboratory limits of detection were 10 mg/kg for REAC SOP 1805, and 5 mg/kg for GC/MS (toluene extraction) and GC/ECD. Limits of detection are based on a 1 gram sample of drywall

The three laboratory methods all showed 100% agreement on detect versus non-detect

for S8. Results from the GC/MS and GC/ECD analyses were generally in good

agreement with each other, although the GC/ECD results were consistently higher than

the GC/MS results for samples with quantifiable S8 concentrations. Although there are

only four detectable samples available for comparison, concentrations of S8 determined

by the REAC SOP 1805 appear less consistent when compared to the other two

laboratory methods.

In addition to the samples measured in this study, Columbia Analytic Services provided

data obtained from analysis of drywall samples in other research not related to CPSC or

EH&E. These additional data points show strong agreement between drywall samples

analyzed by the GC/MS (toluene extraction) and GC/ECD methods (slope = 1.15;

p<0.001) (Figure 4.1), consistent with the finding of strong agreement observed for the

CPSC/EH&E samples presented in Table 4.3.

DRAFT


Figure 4.1 Comparison of S8 Concentrations Measured Using GC/MS and GC/ECD by Columbia Analytical (n=74). The regression line is represented by the solid line and the 1:1 slope is represented by the dashed line.

Determination of S8 concentrations by GC/ECD was also found to be a reasonably

precise method. Results of duplicate analyses (n=6) are presented in Figure 4.2 and

show a 1:1 relationship between duplicate analyses. The mean relative standard

deviation was 23%, and ranged from 1% – 43%, with stronger agreement observed at

the lower end of the concentration distribution.

GC/ECD S8 Concentration (mg/kg)

10 100 1000

GC

/MS

S8

Con

cen

trat

ion

(mg/

kg)

10

100

1000


DRAFT


Figure 4.2 Comparison of S8 Concentrations (mg/kg) in Duplicate Samples as Measured by GC/ECD

4.3.2 Objective 2—Assessment of Intra-board Variability

Multiple measurements of S8 by GC/ECD on the same drywall boards showed very low

intra-board variability (Table 4.4). The average coefficient of variation was 8.5% and

ranged from 0% to 23% across boards. Importantly, the presence or absence of S8 was

consistent across drywall boards. For example, when S8 was determined to be less than

the detection limit in one location, the remaining locations were all also less than the

detection limit. Similarly, if S8 was detected on one location of a drywall board, it was

consistently detected in the remaining locations.

S8 Samples (mg/kg)

0 200 400 600 800 1000 1200

S8 D

uplic

ate

Sa

mpl

es (

mg/

kg)

0

200

400

600

800

1000

1200


DRAFT


Table 4.4 Orthorhombic Sulfur Concentrations (mg/kg) at Multiple Locations Per Board

Sample ID Sample Number

1 2 3 4 5 6 7 8 9 10 11 12 CPSC3 130 130 110 110 120 120 100 110 140 130 150 130CPSC14 <5 <5 <5 <5 <5 <5 <5 <5 <5 <5 <5 <5CPSC32 <5 <5 <5 <5 <5 <5 <5 <5 <5 <5 <5 <5CPSC34 870 700 740 750 700 CPSC35 690 980 1200 980 870 CPSC10 9.9 7.9 7.7 CPSC15 99 96 110 CPSC25 <5 <5 <5 mg/kg milligrams per kilogram

4.3.3 Objective 3—Determination of Orthorhombic Sulfur Concentrations in Catalog Samples and Archived Samples from the 51-Home Study

4.3.3.1 Orthorhombic Sulfur Concentrations in Catalog Samples

S8 concentrations were determined for the full set of catalog samples by GC/ECD. S8

concentrations of Chinese imported drywall boards ranged from <5 to 1,200 mg/kg in the

catalog samples, with a median value of <5 mg/kg (mean = 191 mg/kg). All of the North

American drywall board S8 concentrations were <5 mg/kg. The S8 concentrations

measured by GC/ECD provide the basis for comparing the S8 concentrations against

measurements of gases and corrosion (see Section 5).

4.3.3.2 Orthorhombic Sulfur Concentrations in Archived Samples from the 51-Home Study

In the 51-home study, houses were identified as ‘complaint’ or ‘non-complaint’ based on

homeowner reports to the CPSC that included information on corrosion and odor in the

home, as well as possible health effects (EH&E 2010). House-average S8 concentrations

ranged from <5 to 830 mg/kg in CPSC complaint homes, with a median concentration of

54 mg/kg (mean = 180 mg/kg). For the non-complaint homes, house-average S8

concentrations were all <1 mg/kg. House-average S8 concentrations were significantly

(p<0.01) higher in complaint homes compared to non-complaint homes (Figure 4.3).

DRAFT


Figure 4.3 Comparison of S8 Concentrations Measured In Drywall Samples from Complaint and

Non-complaint Homes in the 51-home Study

Additionally, of the three drywall samples measured for S8 in each of the 10 non-

complaint homes (n=30), none of the individual samples had detectable levels of S8

(detection limit range for individual samples <1 mg/kg – 33 mg/kg; median <10 mg/kg).

An assessment of these S8 results as a marker of gases and corrosion in the home is

presented in Section 6.2, Source Markers and Corrosion.

Complaint Non-complaint

S8 C

once

ntra

tion

(mg/

kg)

0.1

1

10

100

1000

10000

DRAFT


5.0 CHAMBER-BASED CORROSION

5.1 INTRODUCTION

Several methods have been utilized to identify Chinese drywall (e.g., XRF/FTIR, S8,

housing characteristics and corrosion potential). The extent to which these markers

identify all Chinese drywall or, more importantly, only problematic drywall, is unknown.

This section outlines the results of a chamber-based study designed to aid in the

identification of markers of problematic drywall as it relates to a primary dependent

characteristic of problematic drywall, corrosion.

The objectives of this chamber-based study were:

Objective 1

Conduct pilot testing to identify an appropriate chamber testing scenario to

address Objective 2

Objective 2

Identify the potential of drywall in the catalog samples to cause corrosion

5.2 METHODS

During the initial study design phase, the chamber environment was designed to

maintain 77 °F and 50% RH. Preliminary results indicated low rates of corrosion under

these conditions in the test chambers during the eight day exposure period. The

chamber environment was altered, in a second set of tests, to conditions that more

closely reflected outdoor conditions during the 51-home study (90 °F; 50% RH). These

conditions were found to induce corrosion to occur over a time frame of approximately

eight days. The methods for the elevated temperature test are described in detail as

follows. The lower temperature tests were performed in a similar manner.

5.2.1 Chamber Testing

The chamber testing lasted a total of nine days, which included a 23-hour conditioning

phase followed by the silver and copper coupons being exposed for approximately eight

DRAFT


days. Chamber corrosion testing was carried out in a temperature and RH controlled

room (12’ x 9.25’ x 7’, with a 2’ plenum). The room was maintained at 90 °F (32 °C)

+/- 5 °F using a heater and a thermostat, and 50% RH +/- 8% RH using a water bath

filled with deionized water. Air was constantly flowing above the surface of the water

bath. The whole room was kept in a well mixed condition by a large, suspended box fan.

Each test consisted of placing a drywall sample into a 6L stainless steel chamber that

also included a temperature and RH datalogger. An air pump and a charcoal scrubber

were used to condition the stainless steel chamber. The air pump was housed inside the

larger temperature and RH conditioned chamber to avoid condensation in the tubing

(any heat generated by the pumps was shown to not adversely impact temperature

control). The stainless steel chamber was equipped with clamps and a non-volatile

organic compound-emitting Tygon® gasket that allowed the lid to be sealed in place, and

2 threaded Swagelok® ports to permit conditioning of the chamber.

Each test consisted of three phases. In the first phase, a 3-inch by 3-inch gypsum

sample with all four cut edges exposed and the two large faces having intact paper

sheathing was placed on top of a small glass jar inside the stainless steel chamber. The

datalogger was also placed in the chamber at this time. The chamber was then clamped

shut, and the pump, charcoal scrubber, and particle filter were connected to one of the

Swagelok® ports. The other port was left open as an exhaust. For 23 hours the pump

drew temperature and RH controlled air from the chamber, passed it through the

scrubber, and into the stainless steel chamber at a rate of approximately 1 liter per

minute (LPM).

After 23 hours, the test moved into the second phase. The lid of the chamber was briefly

opened and a corrosion classification coupon containing one strip each of pre-cleaned

copper and silver was placed inside, oriented so that the individual metal strips extended

from near the bottom of the chamber to near the top. The coupon was approximately

2.5 inches from the drywall sample. The chamber was again clamped closed, and the

pump continued to deliver conditioned air at 1 LPM (Figure 5.1).

DRAFT


Figure 5.1 Schematic of the Chamber Testing Equipment

After one hour, the test moved to the final phase. The pump, scrubber, and filter

apparatus was disconnected from the Swagelok® port, and both Swagelok® ports were

sealed closed. The chamber was then left in the temperature controlled chamber for

eight days, at which point the chamber was opened, the coupon removed, packaged,

and sent to the laboratory for analysis. The temperature and RH datalogger was also

downloaded and the data checked for temperature or RH variation during the exposure

period. Prior to re-using the system, each chamber was thoroughly cleaned with an

Alconox solution, rinsed with deionized water, and carefully dried.

The corrosion classification coupons were used to determine the integrated corrosion

rate. The corrosion coupons used in this study contained copper and silver metal and

were supplied by Purafil, Inc., Research and Development Laboratory in Doraville,

Georgia. At the end of the sampling period, the corrosion coupons were collected,

placed in sealed containers and returned to Purafil for analysis. The laboratory

measured the thickness of several copper and silver compounds including silver sulfide

(Ag2S), silver chloride (AgCl), Ag ‘unknown’, copper sulfide (Cu2S), copper oxide (CuO),

and Cu ‘unknown’ present in the sample corrosion coupons. The laboratory normalized

the data using the actual period of exposure and reported the result in units of

“angstroms per 30 days of exposure.” This rate is not directly comparable to in-home

DRAFT


measurements because the samples were not taken in ambient conditions. However,

the corrosion rates are directly comparable between samples because they were all

obtained under identical test conditions. Corrosion rates were compared with reference

values contained in the Instrumentation, Systems, and Automation Society (ISA)

Standard ISA-71.04-1985, Environmental Conditions for Process Measurement and

Control Systems: Airborne Contaminants.

5.2.2 Quality Assurance/Quality Control

For corrosion monitoring, blank and duplicate samples were obtained during each round

of testing to estimate the background rate of corrosion for this sampling design.

Additionally, five replicate samples were obtained (Note: duplicate samples refer to

repeat testing using a different piece of drywall from the same larger piece, while

replicate samples refer to repeat testing of the exact same drywall sample that was

originally tested). Temperature and relative humidity were recorded as 5-minute

averages using a HOBO datalogger in each stainless steel test chamber and remained

stable during testing. Over 95% of Temperature and relative humidity readings remained

between 89 and 92 °F, and between 45 and 56% RH, over the 9-day period. Minimum

and maximum temperature and RH recorded were 86 °F, 96 °F, 44%, and 60%

respectively.

Each chamber was given a unique number and samples were randomly assigned to a

specific chamber to reduce the potential for systematic bias. Corrosion rate data were

analyzed for trends by chamber and no systematic bias found.

5.3 RESULTS

The initial study design had a chamber maintained at 77 °F and 50% RH. A total of

13 tests (includes one blank and one duplicate sample) were run under this scenario.

Test results indicated that these chamber conditions were not conducive to generating

measurable corrosion in a short test period (nine days).

Silver and copper corrosion rates were subsequently determined from the final chamber

design (90 °F and 50% RH) for all ‘catalog’ samples and ranged from 94 angstroms per

DRAFT


30 days (A/30d) to 1473 A/30d, and <32 A/30d to 589 A/30d, respectively. The results

from this chamber testing were compared against the S8 and strontium source marker

concentrations, described in the following section of this report (Section 6). Individual

corrosion rate results can be found in the summary table in the Discussion section of this

report (Section 7).

Analysis of blank samples (n=4) showed a low rate of background corrosion. All blank

samples had Cu2S copper corrosion rate of <32 A/30d. For Ag2S, the mean corrosion

rate was 181 A/30d with a standard deviation of 104 A/30d. The method limit of

detection for Ag2S was defined as three times the standard deviation of field blank

results (104*3 = 312 A/30d). Duplicate and replicate tests showed strong agreement and

are presented in Table 5.1.

Table 5.1 Comparison of Corrosion Rates (A/30d) between Sample and Duplicate Samples

Catalog ID Cu2S Ag2S

Sample Duplicate Replicate Sample Duplicate Replicate CPSC2 <32 <32 – 156 109 – CPSC3 227 <32 673 – 592 CPSC5 <32 <32 – 324 302 – CPSC6 <32 – <32 131 – 218 CPSC7 <32 – <32 94 – 128 CPSC8 <32 <32 203 90 – CPSC15 265 – <32 686 – 624 CPSC16 <32 – <32 187 – 125 CPSC18 <32 – <32 842 – 857 CPSC33 <32 – <32 156 – 150 CPSC34 589 – <32 1,473 – 935 CPSC35 530 – <32 1,052 – 1,017 Cu2S copper sulfide Ag2S silver sulfide

DRAFT


6.0 SOURCE MARKERS AND EFFECT

6.1 SOURCE MARKERS AND GAS EMISSIONS

6.1.1 Source Markers and Gas Emissions—Chamber Testing

In a recent report, Lawrence Berkeley National Laboratory (LBNL) reported reactive

sulfur compound emissions data for a number of drywall samples. Most of these drywall

samples overlapped with drywall samples received by EH&E, which were analyzed for

S8 and strontium and provide the basis for comparing gas emissions to source marker

concentrations.

S8 measured in drywall samples in this study was found to be strongly associated with

hydrogen sulfide (H2S) emissions, while strontium levels, taken alone, were generally

found to be a poor predictor of H2S emissions when evaluating all North American and

Chinese samples (Figure 6.1).

Figure 6.1 Scatterplot Showing the Relationship Between a) S8 Concentration (mg/kg) and

Hydrogen Sulfide Flux and b) Strontium (mg/kg) and Hydrogen Sulfide Flux. Open circles represent points where the S8 concentration was <LOD

In addition to hydrogen sulfide, S8 concentrations were also found to be moderately

associated with carbon disulfide (CS2) and sulfur dioxide (SO2) emission rates (Figure

6.2).

GC/ECD: S8 Concentration (mg/kg)

1 10 100 1000 10000

LBN

L: H

ydro

gen

Sul

fide

Flu

x (u

g/m

2/h

)

0

50

100

150

200

250

XRF: Strontium Concentration (mg/kg)

10 100 1000 10000

LBN

L: H

ydro

gen

Sul

fide

Flu

x (u

g/m

2/h

)

0

50

100

150

200

250N = 28Spearman r = 0.72p-value < 0.01

N = 28Spearman r = 0.31p-value = 0.10

DRAFT


Figure 6.2 Scatterplot Showing the Relationship between S8 Concentration (mg/kg) and

a) Sulfur Dioxide Emission Rate and b) Carbon Disulfide Emission Rate. Open circles represent points where the S8 concentration was <LOD.

6.1.2 Source Markers and Gas Emissions—51-Home Study

Consistent with the associations observed between S8 and chamber-based H2S

emissions, S8 and H2S were associated in the 51-home study. House average S8

concentrations in drywall, obtained by averaging the two room composite samples and

one large bulk sample (see Methods for details) were significantly associated with house

average H2S concentrations, controlling for dew point and outdoor hydrogen sulfide

concentrations (Table 6.1).

Table 6.1 Regression Model Results Showing Predictors of House Average Hydrogen Sulfide

Concentrations (Natural log-transformed) in Indoor Air

Parameter Estimate Standard Error p-value Intercept -3.98 0.96 <0.01 Orthorhombic Sulfur (S8) (natural log) 0.08 0.03 0.03 Dew Point 0.05 0.02 0.01 Outdoor Hydrogen Sulfide (H2S) 0.18 0.11 0.09 Model R2=0.37

To explore the potential for interaction, the S8 concentration was dichotomized based on

a cutoff of 10 mg/kg and the regression analyses were re-run. The presence of S8 in

GC/ECD - S8 Concentration (mg/kg)

1 10 100 1000 10000

LBN

L S

O2 E

mis

sio

n R

ate

(ug/

m2/h

)

0

10

20

30

40

50

60

70

GC/ECD - S8 Concentration (mg/kg)

1 10 100 1000 10000

LBN

L C

S2 E

mis

sion

Rat

e (u

g/m

2/h

)

0

1

2

3

4N = 28Spearman r = 0.69p-value < 0.01

N = 28Spearman r = 0.52p-value < 0.01

DRAFT


concentrations greater than 10 mg/kg was significantly associated with house average

hydrogen sulfide, similar to the results using S8 as a continuous variable (p<0.01, model

R2 = 0.41). There was also significant interaction observed between the dichotomized S8

variable and dew point. The effect of dew point on hydrogen sulfide was dependent upon

the presence of S8 (p<0.05).

In similar regression analyses used to identify predictors of H2S, S8 concentrations from

the 51-home study were not associated with CS2 or SO2 concentrations in the homes

(p=0.25, p=0.13, respectively).

For strontium, house average strontium concentrations from the 51-home study were

significantly associated with H2S concentrations in the home. These results differ from

the analysis of catalog samples where strontium concentrations were not associated

with H2S emissions. This difference is discussed in more detail below.

Table 6.2 Regression Model Results Showing Predictors of House Average Hydrogen Sulfide

Concentrations (Natural log-transformed) in Indoor Air

Parameter Estimate Standard Error p-value Intercept -6.28 1.20 <0.001 Strontium (natural log) 0.32 0.13 0.01 Dew Point 0.05 0.02 <0.01 Outdoor Hydrogen Sulfide (H2S) 0.18 0.11 0.10 Model R2=0.38

The evidence presented here suggests that S8 is directly associated with H2S emissions,

and thus is a desirable marker for problematic drywall. Strontium was significantly

associated with H2S in homes from the 51-home study, but not H2S from the chamber

tests of catalog samples. This may be because S8 and strontium are correlated in

problematic drywall, as evidenced in the results from the in-home study (Figure 6.3), but

they are not correlated in the catalog samples, which represent a more diverse cross-

section of drywall samples (e.g., imported, domestic, year of production, etc.). As noted

previously, the strontium marker appears to be useful when used as part of a multifactor

screening process that includes additional information including age of installation and

corrosion potential, among other factors.

DRAFT


Figure 6.3 Correlation of S8 and Strontium in a) Catalog Samples and b) Homes in the 51-home

Study. Open circles represent points where the S8 concentration was <LOD.

6.2 SOURCE MARKERS AND CORROSION

6.2.1 Source Markers and Corrosion—Chamber Testing

A summary of source marker concentrations and corresponding corrosion rates for all

catalog samples is presented in Table 6.3.

Catalog Samples

Strontium Concentration (mg/kg)

102 103 104

S8 C

once

ntra

tion

(mg/

kg)

10-1

100

101

102

103

104

51-Home Study


102 103 104

S8 C

once

ntra

tion

(mg/

kg)

10-1

100

101

102

103

104



DRAFT


Table 6.3 Summary Table

Catalog ID CPSC ID Country of

Origin Strontium

(mg/kg) S8 (mg/kg)Cu2S

(A/30 days) Ag2S

(A/30 days)

CPSC1 09-302-1429-02 Canada 273 <5 <32 150 CPSC2 09-840-9139-05 US 2,580 <5 < 32 156 CPSC3 09-302-1379-09 China – 91 227 673 CPSC4 09-840-9858-01 US 946 <5 <32 118 CPSC5 09-810-7932-05 US – <5 <32 324 CPSC6 09-810-7639-06 US – <5 <32 131 CPSC7 09-840-9961-03 US – <5 <32 94 CPSC8 09-840-9962-08 US – <5 <32 203 CPSC9 09-810-8213-02 US 119 <5 <32 156

CPSC10 09-810-7069-06 China 3,740 7.7 <32 807 CPSC11 09-810-8235-03 US – <5 <32 187 CPSC12 09-810-8036-05 US – <5 <32 193 CPSC13 09-810-8037-01 US – <5 <32 137 CPSC14 09-810-8236-07 US 570 <5 <32 140 CPSC15 09-840-9672-07 China 2,350 99 265 686 CPSC16 09-302-2636-03 China – <5 <32 187 CPSC17 09-840-9707-05 China 351 <5 <32 125 CPSC18 09-840-9673-08 China – 320 <32 842 CPSC19 09-302-1487-02 China 1,500 <5 <32 109 CPSC20 09-302-2634-01a China – <5 <32 125 CPSC21 09-302-1492-02 China – <5 <32 109 CPSC22 09-302-1493-02a China – <5 <32 109 CPSC23 09-302-2631-02b China 5,890 <5 <32 156 CPSC24 09-810-7077-02 China 870 <5 <32 86 CPSC25 09-810-7078-05 China 2,200 <5 <32 312 CPSC26 09-302-2632-01 China 2,720 <5 <32 312 CPSC27 09-302-2633-02 China 2,810 <5 <32 171 CPSC28 09-302-2635-02 China – <5 <32 109 CPSC29 09-840-9667-01 China – <5 <32 249 CPSC30 09-302-2637-02a China – <5 <32 133 CPSC31 09-302-1484-02a US 195 <5 <32 143 CPSC32 09-840-9175-05 US: Used

Imported Materials

6,540 <5 <32 351

CPSC33 09-840-9174-01 US: Used Imported Materials

6,410 <5 <32 156

CPSC34 09-810-7339-10 China – 870 589 1,473 CPSC35 09-810-8357-01 China 273 1,200 530 1,052

CPSC U.S. Consumer Product Safety Commission mg/kg milligrams per kilograms S8 orthorhombic sulfur Cu2S copper sulfide A/30 Angstroms per 30 days Ag2S silver sulfide

DRAFT


S8 concentrations in catalog drywall samples were associated with both silver (r = 0.66,

p<0.01) and copper sulfide corrosion (r = 0.82, p<0.01) as determined in the chamber-

based testing (Figure 6.4).

Figure 6.4 Comparison of S8 Concentrations (mg/kg) and Chamber-based Corrosion Rates

(A/30d). Open circles represent points where the S8 concentration was <LOD.

Strontium concentrations in the catalog samples were associated with copper and silver

corrosion in the chamber testing, but only for catalog samples of drywall that were from

China and produced during the timeframe when problematic drywall was imported

(Figure 6.5).

Figure 6.5 Comparison of Strontium Concentrations (mg/kg) and Chamber-based Corrosion

Rates (A/30d)

S8 Concentration (mg/kg)

100 101 102 103 104C

u2S

Cor

rosi

on R

ate

(A

/30

d)

101

102

103

104

Moderate

Severe

Harsh

Mild

ISA Class


100 101 102 103 104

Ag

2S C

orro

sion

Ra

te (

A/3

0 d

)

101

102

103

104

N=29


101 102 103 104

Cu 2S

Cor

rosi

on

Ra

te (

A/3

0 d

)

101

102

103

104

Moderate

Severe

Harsh

Mild

ISA Class


101 102 103 104

Ag 2S

Cor

rosi

on

Rat

e (

A/3

0 d

)

101

102

103

104

DRAFT


6.2.2 Source Markers and Corrosion—51-Home Study

The house average S8 concentrations were generally associated with corrosion in the

home, as measured by silver and copper corrosion classification coupons at the air

handling unit (AHU) air register (Figure 6.6).

Figure 6.6 Comparison of House-average S8 Concentrations (mg/kg) and Corrosion Rates at

the AHU Air Register from the 51-home Study

Multiple regression analyses showed that S8 concentration (mg/kg; natural log-

transformed) was a significant predictor of silver and copper corrosion rates, controlling

for region, outdoor corrosion rate and indoor temperature (silver – p<0.001, model R2 =

0.56; copper – p<0.001, model R2 = 0.47) (Tables 6.4 and 6.5).

Table 6.4 Regression Model Results Showing Predictors of Silver Corrosion Rate at the AHU

Air Register (Natural log-transformed)

Parameter Estimate Standard Error p-value Intercept 5.23 1.25 <0.001 Orthorhombic Sulfur (natural log) 0.26 0.04 <0.001 Region – Florida East 1.46 0.35 <0.001 Region – Florida West 0.97 0.37 0.01 Region – Gulf Coast 0.68 0.34 0.05 Region – Virginia -- -- -- Outdoor Silver Sulfide (Ag2S) <0.001 <0.001 0.20 Dew Point <0.001 0.02 0.99 AHU air handling unit Model R2=0.56


10-1 100 101 102 103 104

Ag

2S

Cor

rosi

on R

ate

(A

/30

d) a

t AH

U A

ir R

egis

ter

101

102

103

104


10-1 100 101 102 103 104

Cu

2S

Cor

rosi

on R

ate

(A/3

0 d)

at A

HU

Air

Reg

iste

r

101

102

103

104

Moderate

Severe

Harsh

Mild

ISA Class

DRAFT


Table 6.5 Regression Model Results Showing Predictors of Copper Corrosion Rate at the

AHU Air Register (Natural log-transformed)

Parameter Estimate Standard Error p-value Intercept 3.94 2.60 0.1367 Orthorhombic Sulfur (natural log) 0.42 0.09 <0.001 Region – Florida East 2.25 0.76 <0.01 Region – Florida West 1.59 0.74 0.04 Region – Gulf Coast 1.15 0.70 0.11 Region – Virginia -- -- -- Outdoor Copper Sulfide (Cu2S) <0.001 <0.001 0.05 Dew Point -0.021 0.04 0.62 AHU air handling unit -- Referent group Model R2=0.47

When the S8 concentration was dichotomized based on a cutoff of 10 mg/kg this

relationship remained, with the S8 marker accounting for 49% and 46% of the variance in

silver and copper corrosion, respectively. No significant interaction was found for the S8

marker and environmental parameters such as temperature and relative humidity.

Analysis using the house average strontium concentrations yielded similar results.

Strontium was significantly associated with both silver and copper corrosion at the AHU

air register (p<0.01, R2=0.65; p<0.01, R2=0.46) (Figure 6.7).


102 103 104

Ag 2

S C

orro

sion

Rat

e (A

/30

d)

at A

HU

Air

Reg

iste

r

101

102

103

104


102 103 104

Cu 2

S C

orr

osi

on

Ra

te (

A/3

0 d

) at

AH

U A

ir R

egis

ter

101

102

103

104

Moderate

Severe

Harsh

Mild

ISA Class

Figure 6.7 Scatterplot Comparing Strontium Concentrations (mg/kg) and Silver and Copper

Corrosion Rates (A/30 days) at the AHU Air Register

DRAFT


7.0 DISCUSSION

7.1 OVERVIEW

In an earlier study involving 51-homes (complaint and non-complaint) XRF and FTIR

were evaluated as methods for identifying markers of imported drywall (EH&E 2010).

Strontium and carbonate were identified as two markers that, if analyzed together, were

useful in identifying imported drywall (EH&E 2010). In this previous study, homes with

the carbonate and strontium marker determined by FTIR and XRF, respectively, were

found to have significantly elevated concentrations of H2S and corrosion compared to

homes without the marker present. While the 51-home study was being conducted, an

allotrope of S8 in the drywall was suggested as another potential marker of problematic

drywall. Because of the critical need to have accurate means of screening for

problematic drywall potentially impacted homes, a series of experiments were proposed

to provide further insight. The objectives of this study, designed to provide information

relevant to this inquiry, were to:

1. Determine the precision and accuracy of strontium concentration measurements in

drywall made with field portable instruments.

2. Determine S8 content in drywall samples from the CPSC inventory (‘catalog drywall

samples’) and drywall samples archived from the 51-home study.

3. Characterize the potential of catalog samples to cause corrosion.

4. Identify source markers of problematic drywall by comparison of source marker

concentrations to both chamber-based and in-home measurements of gases and

corrosion.

7.2 STRONTIUM CONCENTRATIONS USING PORTABLE XRF ANALYZERS

Four different XRF units (three manufacturers) were compared to assess the

consistency of results. The strontium concentrations measured with the four instruments

were very highly correlated (r>0.99, p<0001). If the instruments were calibrated with a

standard reference material containing the element being measured, the results were

essentially interchangeable. XRF results were also compared to ICP-AES

determinations of strontium. In paired samples the strontium concentrations measured

by XRF were very consistent with concentrations from the ICP-AES analyses

DRAFT


(r 0.85-0.95, p<0.01). This finding indicates that portable XRF analyzers can be reliably

used to quantify strontium concentrations.

7.3 COMPARISON OF ANALYTICAL METHODS FOR ORTHORHOMBIC SULFUR

Three methods for analyzing S8 were used to measure S8 concentrations in samples of

drywall. These drywall samples were obtained from known sources and contained

products that were both domestically produced and, imported. The methods were REAC

SOP 1805, an EPA method; GC/MS with a toluene-based solvent extraction, and

GC/ECD. The GC/MS and GC/ECD methods generally were in agreement but the REAC

SOP 1805 was not as consistent when compared to the results from the other two

methods. There was 100% agreement between all three methods when the criteria was

detect vs. non-detect for S8. The GC/ECD method was selected as the method for

analyzing S8 in the remaining catalog and in-home samples in this study. The GC/ECD

method showed strong internal consistency, has been demonstrated to be useful for

drywall analysis (Singhvi et al. 2009), and has a lower cost than the other two methods

for laboratories with this capability.

7.4 INTRA-BOARD VARIABILITY OF STRONTIUM AND ORTHORHOMBIC SULFUR CONCENTRATIONS

When multiple measurements of strontium and S8 were made on the same drywall

board, there was low intra-board variability. Importantly, the presence or absence of S8

was consistent across drywall boards. For example, when a S8 was determined to be

less than the detection limit in one location, the remaining locations were all also less

than the detection limit. Similarly, if S8 was detected on one location of a drywall board, it

was consistently detected in the remaining locations. The results of this portion of the

study indicate that there is little variability in sampling results by selecting different parts

of the board, a finding that is consistent with previously reported results (Alessandroni,

2009). This should make in-home/field sampling more straightforward and efficient and

increase confidence in results obtained with a few number of samples.

DRAFT


7.5 EXPOSURE PATHWAY: SOURCE – EXPOSURE – EFFECT

Multiple lines of evidence were pursued to evaluate the robustness of associations

between source markers (i.e., strontium and S8), exposure (i.e., gases), and effects (i.e.,

corrosion). This included assessing the relationships observed in controlled chamber

studies, and also associations observed in homes.

Several of the drywall samples that had been analyzed by LBNL for reactive sulfur gases

were from the same source sample materials that EH&E had analyzed for S8 and

strontium. This subset of drywall samples and analytical data provide the basis for

comparing gas emissions to source marker concentrations (sulfur and strontium). In this

study S8 measured in drywall samples was found to be strongly associated with H2S

emissions, while strontium was found to be a good predictor of H2S emissions in drywall

samples suspected of being problematic due to both their origin and date of production.

The results of the in-home study indicate that both strontium and S8 were both significant

predictors of H2S concentrations in homes. These homes were suspected of containing

problematic drywall due to reports of odors, corrosion, health complaints, and,

importantly, meeting the criteria of being constructed during the time when problematic

drywall was imported. The comparison of markers and H2S emissions in these samples

indicates that S8 is a reliable predictor of H2S emissions regardless of any pre-screening

criteria. However, strontium measurements can be obtained rapidly, non-destructively,

and in situ. Therefore, in-home screening conducted using XRF and analyzing for

strontium is a useful tool in identifying problematic drywall provided that pre-screening

criteria are met (e.g., odor, corrosion, health complaints). Confirmatory analysis using S8

as a marker could be used to verify the results.

Strontium and S8 concentration in drywall samples were also found to be associated with

both chamber-based corrosion rates and corrosion rates measured in homes. Stronger

associations were observed for Ag2S corrosion compared to Cu2S corrosion. One

possible explanation for this observation is related to the mass increase rates. Copper

and silver both have fast mass increase rates, but, in an experimental study, silver was

shown to have a mass increase rate an order of magnitude faster than copper, showing

mass responses within several minutes after H2S was introduced (Forslund et al. 1997).

DRAFT


In this study’s chamber-based corrosion tests, corrosion was strongly associated with

the sulfur concentration; however, the strontium concentration alone was found to be a

poor predictor of corrosion (as measured in the chambers) without any information on

the origin of the drywall samples. The lack of association with corrosion for strontium

differs from that in the 51-home study. In that study strontium was associated with

problem homes and corrosion in those homes. This apparent inconsistency may be the

result of the aforementioned selection process for homes in the 51-home study. In the

51-home study, the homes were selected based on several criteria that helped identify

them as ‘problem’ homes. In samples of drywall from those homes, strontium and S8

were highly correlated, As a result, they were both useful markers of problematic drywall

for homes in that study. The catalog samples tested in this study contain a diverse mix of

domestic and imported drywall, some manufactured during the period of prime interest

and others after, and some of which is problematic and some of which is not. For this

wide range of samples, without use of any additional characteristics, strontium was not a

consistent predictor of problematic drywall.

7.6 SUMMARY

In the 51-home study, indoor H2S levels were found to be associated with corrosion and

the home being classified as a “problem home” (EH&E 2010). Although the mechanism

responsible for the corrosion has not been elucidated, the strong relationship observed

in this study between sulfur content of problem drywall and H2S emissions is supportive

of S8 being a reliable marker for problematic drywall. Further support for this conclusion

is the finding that concentration of sulfur in the drywall is associated with corrosion in

testing chambers and that in multiple regression analyses of corrosion measured in

homes during the 51-home study, sulfur concentration in drywall was significantly

associated with corrosion rate (while controlling for temperature, humidity, and region).

This consistency with experimental studies, observations in the field and statistical

analyses which control for potential effect modifiers/confounders, supports the use of S8

in drywall as a reliable indicator of problematic drywall. In addition, S8 and strontium

concentrations were found to be correlated in problematic drywall and strontium was an

equally strong predictor of H2S and corrosion in problematic homes. Therefore, the

ability to test for strontium concentration in the field instantaneously and non-

DRAFT


destructively using portable XRF analyzers suggests that strontium is useful as a

screening tool for problematic drywall when pre-screening criteria are utilized.

DRAFT


8.0 CONCLUSIONS

The objective of the source characterization study was to evaluate proposed markers of

problematic drywall, defined as drywall associated with elevated rates of copper and

silver corrosion. In this study, we investigated two proposed markers—strontium and S8.

The robustness of each source marker was evaluated by first comparing different

instruments/methods for quantifying each, examining parameters such as within-board

variability and determining method precision. The source markers were then compared

to both chamber-based and field-based measurements of gases and corrosion. The

following observations were made during the course of this investigation:

Strontium is a useful, but non-specific marker of problematic drywall when used in

isolation

Elevated strontium concentrations were observed in all problematic drywall, but

also in some non-problematic drywall. Strontium concentrations were correlated

with S8 concentrations in problematic drywall. Therefore, in the 51-home study

where homes were pre-screened based on specific criteria contained in a CPSC

questionnaire (EH&E 2010) strontium was found to be predictive of problematic

drywall. Strontium content in drywall measured by XRF is non-destructive, field

portable and nearly instantaneous, and, therefore, remains a useful marker of

problematic drywall when used as part of a multi-level screening approach.

Analysis of strontium in drywall samples can be reliably performed using XRF

Strontium measurements using XRF were accurate when compared to strontium

as determined by ICP-AES (inductively coupled plasma-atomic emission

spectroscopy). Models from three different manufacturers yielded highly similar

response factors and strongly correlated results (1:1 slope). Instrument method

calibration specific to strontium is necessary to ensure accuracy of the

measurements.

S8 is a sensitive and specific marker of problematic drywall

S8 concentrations in drywall were associated with chamber-based measurements

of hydrogen sulfide and corrosion. Consistent findings were observed when this

DRAFT


relationship was evaluated using archived samples of drywall and measurements

of hydrogen sulfide and corrosion in the 51-home study.

S8 was not detected in any drywall samples from the non-complaint homes in the 51-

home study

Three drywall samples from each of the 10 non-complaint homes in the 51-home

study were analyzed for orthorhombic sulfur. S8 was not detected in any of these

samples. In contrast, S8 concentrations in the complaint homes ranged from

<5 mg/kg to 830 mg/kg (median = 54 mg/kg), and were significantly higher than

the levels in the non-complaint homes.

S8 determined using two toluene-based extraction methods showed strong

agreement

S8 concentrations determined using the GC/MS (toluene extraction) and GC/ECD

methods showed excellent agreement. Results using EPA’s REAC SOP 1805 did

not show consistent agreement when compared with the other two methods, in a

limited number of samples. GC/ECD appears to be an attractive option for future

analysis of drywall samples due to its potentially lower cost for laboratories with

this capability.

S8 and strontium both exhibited low intra-board variability

Repeat measurements of S8 and strontium on different locations of the same

drywall board showed strong consistency.

DRAFT


9.0 REFERENCES

Alessandroni, M.2009. What's the (Elemental S)tory? Poster presented at the Technical Symposium on Corrosive Imported Drywall. Tampa, Florida. November 5-6, 2009. EH&E. 2010. Final Report on an Indoor Environmental Quality Assessment of Residences Containing Chinese Drywall. Needham, MA: Environmental Health & Engineering, Inc. January 28, 2010. EPA. 2009a. Drywall Investigations: Additional Five Drywall Sample Analysis Summary Results. U.S. Environmental Protection Agency, Environmental Response Team Memorandum dated August 27, 2009. EPA. 2009b. Drywall Investigations: Fifteen CPSC Drywall Sample Analysis Summary Results. U.S. Environmental Protection Agency, Environmental Response Team Memorandum dated September 16, 2009. EPA. 2009c. Drywall Sampling Analysis. U.S. Environmental Protection Agency, Environmental Response Team Memorandum dated May 7, 2009. Forslund M, Majoros J and Leygraf C. 1997. A sensor system for high resolution in situ atmospheric corrosivity monitoring in field environments. Journal of the Electrochemical Society. 144(8):2637-2642. ISA. 1985. Environmental Conditions for Process Measurement and Control Systems: Airborne Contaminants. ISA Standard ISA-71.04-1985. Research Triangle Park, NC. Instrumentation, Systems, and Automation Society. Singhvi R, Lin Y, Admassu G, and Syslo J. 2009. Field Analysis of Elemental Sulfur in Drywall by GC/ECD. Poster presented at the Technical Symposium on Corrosive Imported Drywall. Tampa, Florida. November 5-6, 2009.

Identification of Problematic Drywall: Source Markers and ...Contract Standard Operating Procedure...

Documents

Transcript of Identification of Problematic Drywall: Source Markers and ...Contract Standard Operating Procedure...